Kuingsmile/word-GPT-Plus
Word GPT Plus is a word add-in which integrates the AI&Agent into Microsoft Word.
Word GPT Plus – AI‑powered assistant inside Microsoft Word
What it is – A Microsoft Word add‑in that lets you talk to large‑language‑model services (OpenAI, Azure OpenAI, Google Gemini, Groq, Ollama, …) from the Word UI. The add‑in can act as a simple chat bot or as an agent that runs tools to read, write and format the current .docx document.
Key capabilities
- Chat mode – one‑off Q&A, text generation, translation, polishing, summarisation, grammar checking.
- Agent mode (built on LangChain) – the model can call over 25 built‑in Word‑specific tools: insert/replace text, create tables, apply styles, search‑and‑replace, web search, calculations, etc. The agent’s reasoning steps are shown in a collapsible view and responses stream live.
- Multi‑model support – plug‑in works with any OpenAI‑compatible endpoint, Azure OpenAI deployments, Google Gemini models, Groq models, and local Ollama servers. Custom model names, temperature, token limits and base URLs are configurable.
- Quick actions – one‑click buttons for translation (40+ languages), polishing, academic‑style improvement, summarisation and grammar correction.
- Formatting – AI output is automatically converted to proper Word styles; markdown in the response is parsed into Word formatting.
- Privacy – API keys and prompts are stored only in the browser’s local storage inside the add‑in; the add‑in talks directly to the chosen AI provider.
How to get it
- Instant‑use – download the provided
manifest.xml, place it in a folder and sideload the add‑in following Microsoft’s Office‑add‑in guide. - Self‑hosted – run the official Docker image or build from source (Node 24 LTS, pnpm 12). After deployment, edit the manifest to point at your server.
- EdgeOne deployment – one‑click deploy to Tencent EdgeOne (Node 24 required).
Prerequisites
- Microsoft Word 2016/2019/2021 or Microsoft 365 (only .docx files).
- Edge WebView2 Runtime.
- API keys for the chosen LLM provider.
Typical workflow
- Open Word, load the GPT Plus pane.
- Enter your API key in Settings → Provider.
- Choose Chat for a quick rewrite or Agent to let the AI manipulate the document (e.g., “Summarise the whole report and insert the summary at the top”).
- Watch the streamed response and, if needed, abort or edit the result.
Who it’s for – writers, researchers, students, or anyone who wants AI assistance without leaving Word. It’s especially handy for repetitive editing, generating drafts, or extracting information from long documents.
License – MIT.
Related
- Project
- Project
- Dispatch
- Project